JOURNAL O F T H E AMERICAN CHEMICAL SOCIETY Krgi.stered in L'.S. Pafrnt Office. 0 C'opj,righf, I980, hi' rhe Aniwiruti Cheniical S o c i e f ~ .
V O L U M E 102, N U M B E R3
J A N U A R Y30, 1980
An NMR Study of the Structure of the Antibiotic Ristocetin A. The Negative Nuclear Overhauser Effect in Structure Elucidation John R. Kalman and Dudley H. Williams" Contribution f r o m the Unicersity Chemical Laboratory, Cambridge, CB2 I E W , United Kingdom. Receired Februarj, 22, 1979
Abstract: An analysis of the ' H N M R spectrum of ristocetin A in dimethyl sulfoxide solution is reported. With the aid of negative nuclear Overhauser effects, and analogies between a part of the structure of the anitbiotic and a part of that of vancomycin, it has been possible to propose a structure for ristocetin A aglycone. The study leads to proposals for the absolute stereochemistry at eight of the nine asymmetric centers. Some details of the attachment of the six sugars of ristocetin A (previously known to be attached as two monosaccharides-ristosamine and mannose-and as a tetrasaccharide via glucose) are deduced.
Introduction Recent work has established that the antibiotic ristocetin A (identical with ristomycin AI) is constituted from six suga r ~ *(ristosamine, -~ mannose ( 2 mol), glucose, arabinose, and rhamnose), and aromatic fragments 1,5 2,1,6and 3.537The production of glycine upon alkaline hydrolysis of ristocetin A, in conjection with evidence from I3C and ' H N M R data, and a californium plasma desorption mass spectral determination of the molecular weight (to within 5 daltons) lead to the conclusion' that (like vancomycin) ristocetin A contains two
"potential glycine" units, 4 and 4a. The IH N M R data further establish the presence of six N H groups of secondary amide bonds.] Since ristocetin A is known to contain two free amino groups (one in the amino sugar ristosamine) and a methyl ester function,8 the molecular weight and N M R data are satisfied by connection of the units 1 to 4 by secondary amide bonds. The six sugars are reported to be attached to the aglycone as a tetrasaccharide (comprising rhamnose, arabinose, mannose, and glucose) and two monosaccharides (ristosamine and m a n n ~ s e )The . ~ parts of a total aglycone structure 5 which
H
0002-7863/80/ 1502-0897$01 . O O / O
0 1980 American Chemical Society
891
Journal of the Anieriran Cheniical Society
898
Table 1. N M R Parameters of Ristocetin A at 70 " C (270 MHz) chemical shift,
d
protons decoupled ( J ) on irradiation
intenSlt)
(multiplicit))
9.26 8.58 7.86 7.66 7.55 7.41 7.4 1 7.29 7.26 7.25 7.21 7.21 7.20 7.18 7.12 7.03 6.88 6.85 6.84 6.77 6.59 6.45 6.42 6.32 5.85 5.65 5.49 5.38 5.33 5.28 5.25 5.19 5.17 5.09 4.92 4.88 4.83 4.73 4.51 4.55 4.38 3.92.8 2.21 2.0 1 2.01 1.25 1.10
letter code
J,
Hz 5 6.5 8.0 8.2 8.2
-8 -IO -8 -2 -8 -8, -2 -9 -I 2 -8, -2 -8. -2 -8 8, -2 2 8. -2 8
8.2
a5
a6 h I
Studies of Ristocetin A in MezSO Solution Examination of spectra taken at a variety of temperatures between 24 and 70 OC with or without added trifluoroacetic-dl acid or DzO, with the help of decoupling experiments at 50 and 70 OC enables us to list the protons in the chemical shift ranges 6 10-4.2 and 2.8-0 (Table I ) . Most protons in the 6 3-4 region are unassigned. The 270-MHz IH NMR spectrum of ristocetin A (1 2 m M in MezSO-d6 at 70 "C) is reproduced in Figure 1 ; expansions and assignments of the 6 4.0-6.0 and 6.2-8.0 ppm regions are given in Figures 2 and 3, respectively.
j m
I k n P 9 r t s3
I v, $3 a3
X
s5
y s4
aa bb cc s2
dd SI
s6
sugar C H and O H ; 15 eel; bb gg
-5 6
hh
[ b 3.461 16 3.311
have been inferred from spectroscopic evidence' are indicated b j dotted lines. We outline in the present paper reasons for the particular connectivity scheme of 5 and almost complete structural details for ristocetin A. sugar glucose ristosamine mannosel
Experimental Section Solutions (0.5 mL) of ristocetin A (supplied by Abbott, Chicago,
The spectra were obtained using Bruker WH270 or 360 spectrometers equipped with Fourier transform facilities. In a typical experiment, 100 scans were accumulated over a spectral width of 3000 Hz using 8 K memory points: 0.4-Hz exponential line broadening was introduced on transformation. Double irradiations for decoupling experiments were performed by continuous irradiation at a power attenuation of 20 dB. The NOES were measured either from these spectra or from spectra recorded using a gated irradiation pulse sequence, by comparison of peak heights of several spectra obtained under closely similar conditions.
a4
d f e E
z
5 12
mol-'). What follows is a systematic study of ristocetin A and (in the following paper) of the complex by IH N M R at 270 and 360 MHz.
OC.
W
6.5
Januar), 30, I980
C
U
9.5
/
or Lundbeck, Copenhagen) in tubes of 5-mm 0.d. were examined at concentrations around 12 mM in MezSO-db and in MezSO-dh containing added D2O (-7%) or added CF3COOD (up to ca. 40 mM). Spectra were recorded at several temperatures ranging from 23 to 70
Y
7.8 IO -5
102:3
a1 a: a a3 b
0
2
1
Code for Anomeric Protons code sugar X mannose2 bb rhamnose M. arabinose
code aa dd U
It has previously been establishedIO that ristocetin A binds
to acetyl-D-Ala-D-Ala, which is a model for a component involved in a key step of cell wall biosynthesis." Moreover, we have already shown' that acetyl-D-Ala-D-Ala forms a strong complex with ristocetin A in Me2SO-d6 solfition. This complex is in slow exchange with its free components on the NMR time scale (calculated free-energy barrier for dissociation -1 8 kcal
Assignment of the Signals A. Phenolic OH Protons. Addition of 0.3 molar equiv of trifluoroacetic-dl acid (TFA-dl) causes the broad phenolic protons in the 6 9-10 region to sharpen; four phenolic OH protons are discerned a t 6 values of 9.73,9.69,9.57, and 9.39 ppm. These resonances are removed on addition of D l O to the MezSO-ds solution, and lose intensity upon irradiation of the HODpeak. B. Amine NH2 Protons. Addition of 0.7 molar equiv of TFA-d, causes a three- to four-proton singlet to appear at 8.5 1 ppm, and also causes similar increase in integrated proton intensity at about 7.6 ppm. These resonances cannot be detected a t higher pH owing to intermediate exchange rates. These signals correspond to two -+NH3 resonances. C. Aliphatic OH Protons. A large number of aliphatic OH resonances are present in the region 6-3 ppm. These are distinguished from other resonances i n the same region by broadening and upfield shifts with increasing temperature, and by pH-dependent line width. No attempt has been made to give detailed assignments of the aliphatic O H protons. D. Protons of Secondary Amide Groups (-CONH-). There are unambiguously six secondary amides in ristocetin A, resonances which ( I ) disappear upon D 2 0 exchange but are slow to exchange on the N M R time scale and (2) are coupled to a C H protons in the 6-4-ppm region. The secondary amide N H protons are coded a ] , a2, . . . , a6 in Table I. E. Aromatic Protons. There are 20 aromatic protons in ristocetin A . Of these, 18 resonances (coded a, b, c, . . . r ) occur in the region 8.0-6.3 ppm and two a t higher field (coded t and v), The aromatic protons are identified by their characteristic chemical shifts, their mututal couplings with standard values of Jortho 8 and J,,,, 2 Hz, and lack of exchange on addition of D20. The two high-field aromatic protons are distinguished from anomeric protons, a - C H , and C H O H protons in being mutually coupled but lacking a coupling to an amide N H . The chemical shifts of the protons t and v are consistent with values already reported for vancomycinI2 (which contains
-
-
Kalnzan, Williams
/
Structure of Antibiotic Ristocetin A
I
I
10
9
I 0
I 7
I 6
899
I 5
I 4
I
I
2
3
I
I
1
0
Figure 1. 70-MHz ' H V M R spectrum of ristocetin A (12 mM in MezSO-ds at 70 "C).
I
t
dd
A
A
r
12
6,O
5,6
5,2
4,8
4,4
4,O
Figure 2. Expansion and assignment of the d 4.0-6.0 ppm region in the 270-MHz ' H N M R spectrum of ristocetin A (12 mM in Me2SO-d6at 70 "C).
a three-ring fragment similar to 1). The high-field chemical shifts o f t and v establish that these protons lie under rings I and 111, Le., the planes of these rings are approximately perpendicular to the plane of ring I1 as in v a n c ~ m y c i n . ' ~ , ' ~ F. The aCHNH2 Proton. Only a single resonance (coded cc) due to a carbon-bound proton in the expected chemical shift .range 6-4 ppm shows a strong shift dependence on the acidity of the solvent. For example, upon addition of 1 molar equiv of T F A to ristocetin A in MezSO-d6 a t 30 OC, a resonance (cc) initially appearing a t 4.87 ppm shifts downfield to 5.47 PPm, G. ArCH(0R)CH- Protons. The a - C H resonance s4 is a doublet of doublets, being not only coupled to an N H proton
(as, 7.21 ppm, J = 9 Hz) but also to resonance y (5.19 ppm, J = 5 Hz). These data define the first -CH(OR)CHNHgroup. Irradiation of the a C H resonance s6 results in a perceptible sharpening of the broad singlet z (5.17 ppm), although the S ~ / Zmutual coupling is small (-1 Hz); thus the assignment of the second -CH(OR)CHNH- group is established. Further coupling of y and/or z to an aliphatic O H was demonstrated in solutions of suitable pH by irradiation of the superimposed doublet and broad singlet O H protons resonating at -5.98 ppm. These data demonstrate two -CH(OR)CHNH- groups in ristocetin A, wherein at least one group R = H. H. Sugar Protons. Most of the sugar C H resonances occur in the region 2.5-4 ppm, but the six anomeric protons resonate
900
Journal of the American Chemical Society
/ 102.3 /
January 30, 1980
D
II
I
‘ I
nf
0
II
II
r
n
A
~~
8,O
7,6
6,8
7,2
~
~~
6,4
Figure 3. Expansion and assignment of the d 6.2-8.0 ppm region in the 270-MHz ‘ H N M R spectrum of ristocetin A ( 1 2 mM in MezSO-d6 at 70 “C). Resolvable mutual couplings are indicated by straight lines; meta-coupled pairs of resonances are indicated by matching symbols below the spectrum
in the 4-6-ppm region. The anomeric protons will be nonexchangeable and coupled, if a t all, only to protons in the 2.54-ppm region (excepting the anomeric proton of the 2-deoxy sugar ristosamine). Of the six thus far unassigned resonances in the 4-6-ppm region, there is only a single doublet (x, 5.28 ppm); this is decoupled upon irradiation a t 3.70 ppm. Of the remaining five (u, w, aa, bb, and dd), one resonance (bb) exhibits coupling to a one-proton resonance (ee2) a t 2.01 ppm; ee2 is geminally coupled ( J = 1 5 Hz) to eel (2.21 ppm). Resonance bb is thus assigned to the ristosamine anomeric proton (see 6 ) . Resonances u, w, aa, and dd are all singlets, and none has been successfully decoupled from a resonance in the 2.5-4-ppm region. These anomeric protons do not therefore manifest large (e.g., diaxial) couplings with C-2 protons; the stereochemical information available from these observations is discussed subsequently. The doublet methyl resonances at 1.25 (gg) and 1.10 ppm (hh) are coupled to protons at 3.46 and 3.31 ppm in the “sugar envelope”, consistent with their being those of the 6-methyl groups of ristosamine and rhamnose. The protons of the aromatic methyl group of 3 resonate a t 2.01 ppm (ff). Specific Assignment of the Ristocetin A Spectrum in Me2SO-& The assignments considered so far, and many more to be detailed subsequently, are based on (1) decoupling experiments, (2) nuclear Overhauser effects (NOEs), and (3) studies of the Ac-D-Ala-D-Ala/ristocetin A complex (to be considered in more detail in the following paperI4). Aromatic Protons. These are grouped on the basis of decoupling experiments. The meta-coupled pairs are r l; f
+ +
+
+
+
+
k; b c: h 0 ; a i, g t m; and t v. From the coupling patterns, the assignments that follow for rings 1-111 are given in 7-9.
6
7
It would normally be expected that both the protons ortho to the CH group would be at lower field than those ortho to oxygen. Whereas this is the case for ring I (see 7 and below), it is not the case for ring I11 (9) since a and i are meta coupled. Confirmatory evidence for the above assignments, and further information, is available from the observation of nuclear Overhauser effects (NOEs) (Table 11). These NOEs, and others detailed subsequently, are observed during the course of normal spin decoupling experiments; they are observed at a variety of temperatures. Since (1) ristocetin A has a relatively high molecular weight, and therefore tumbles in solution relatively slowly, and (2) our spectra have been obtained a t high field
Kalman, Williams
/ Structure of Antibiotic Ristocetin A
strengths (270- and 360-MHz spectra), these NOEs are negative in sign.'* That is, when two protons are sufficiently close in space that one contributes to dipolar relaxation of the other, irradiation of one causes a decrease in the intensity of the resonance of the other. These observations allow crucial deductions to be made as to the molecular geometry of ristocetin A.
The N O E data are interpreted i n terms of close proximity of the protons where the letter codes are joined by doubleheaded arrows in 10.
/
n
90 1 Table 11. h O E Data Relekant to Assignments in Fragments 7-9 intensity reduced
protons irrad d
(%)
+ a6
t ( 1 5%), v a (30%)
g + a5
'+
2
(12%)
b (15%),
x+y+
(a5 + E)" i (20%) I (20-50%)
S S + X + )
(%I
b ss
a
intensity reduced
protons irrad
+z
I
(8 + h ) " d (20%) b (20%), z ( 10%)
V
t s6
Overlapping resonances
relevant data are given in 17-19; the ristocetin A data (17 and 18) refer to a temperature of 40 O C and the vancomycin data
\/ \ /
;+ 11
12
14
13
C'H'
C'H'
r or I 15
I
CH t 16
These restrictions have been confirmed by N O E experiments performed on the Ac-D-Ala-D-Ala complex of ristocetin A.I4 The four phenolic protons which are evident in the spectrum of the complex were irradiated in turn. Resonances corresponding to r and 1 showed negative NOEs on irradiation of a single phenolic resonance. Therefore r and 1 must both be ortho to the same phenolic group, as in 15 (Rl = H). Thus the oxygen substituent between r and I does not carry a sugar residue and is indeed a phenolic OH group. In sequential experiments, irradiation of two other phenolic protons led to negative NOEs for resonances correspong to n and p. Hence both n and p are ortho to phenolic groups. It follows that in 16, R = H and that the proton ortho to the OH group is p, and that the group R in 11 is OH. In 15, R2 may be a sugar residue (as is established subsequently). We have established that the attachment of 11 to 13 constitutes the biphenyl system (rather than 12 to 13) by comparing the observed chemical shifts with those found in the same biphenyl unit when incorporated in vancomycin.12 The
CH
\ / CH
\ /
q -Q-
suggests that the aromatic methyl group is present on the same ring as p and q, and advances the assignments to those shown in 15 and 16.
\/
CH I
I
In addition, two 1,2,4-trisubstituted rings (11 and 12) and two 1,2,3,5-tetrasubstitutedrings (13 and 14) can be characterized. Both p and q are broad resonances since the meta coupling cannot be resolved, whereas the meta coupling is readily resolved for the resonances 1 and r. This observation
k\ n+
\/
CH
I
19
to a temperature of 70 "C. There is a close correspondence of chemical shifts in the biphenyl units of ristocetin A (17) and vancomycin (19) with the exception of resonance I. This one major different is not unexpected, since it will be shown that the oxygen atom ortho to both 1 and the inter-ring bond of the biphenyl system carries a sugar in ristocetin A, but carries a hydrogen atom in vancomycin.I3 Ring Connectivity. Assignment of aCH/NH Pairs and CHOH Groups Nuclear Overhauser effects, which aid the proton assignments and structure elucidation, are listed in Table 111. I n developing the connectivity arguments which follow from NOEs, we will refer to the assigned aromatic ring systems of 10, 17, and 18 by the Roman numerals I-VI1 indicated in these structures. On the basis of strong NOEs, we conclude that the following proton pairs are in close proximity: (a as) (a3 t) (P S 5 ) (a4 4) (a5 9) (s6 f) (s6 S2) (f S2) (b 2 ) (i Y). The 6-hydroxytyrosine units [-CH(OH)CH(NH)-] are associated with the resonances y, s4, and a5 and z, s6, and a6. The former set can be placed on ring 111 by the proximity of a and a5, the latter set on ring I by the proximity of b, s6, and z (see 10). Irradiation o f t causes s3 to sharpen slightly, placing s3 on ring 11. The N O E s2 f places s2 on rings VI or VII, and p s5 places s5 on ring V (or, less probably, on ring IV). The mutual NOEs of s2 and s6 (and f) indicate that ring I is connected to the ring bearing s2 (i.e., ring VI or VII). An N O E (c f a6) S I , observed in the spectrum of the peptide-antibiotic complex,14shows that s' is also in this region of the molecule. It follows that the N O E (e f) S I must be due to the proximity o f f and S I (since e is in ring 111). We
- --
-- - - -
-
+ +
-
+
-
Journal of the American Chemical Society
902
/
102.3
/
January 30, 1980
Table 111. NOEs for Ristocetin A in MezSO-db Solution under Various Conditions" proton irrad
i n t reduced (%)
proton irrad
int reduced (%)
+
a
h (20%); a5 g t (20%) 2 (>12%) q (lo%), 0 ( I 2%) t ("I 5%) t ( 1 3%), s2 (25%) 51 (-20%) a (30%) w (25%),~(10%) i (>20%) s6 (40%), f (30%) ~2 (60%). f (25%) b (20%). z (10%) h (15%)
a3
b c a4 d a6 e+f
+ +
g + a5
k+l+m x+y+sg s2
s6
aa
+ bb + cc
s2
(-20%), cc ( I 3%)
a b ( 1 5%) t, P (40%) 9 (23%) Z
("1
o%), S 2 ( 14%)
w (30%) s2
(20%)
a3 sg
53
(>20%) (25%)
a The relevant spectra were recorded at a variety of temperatures between 24 and 70 ' C . As a consequence, near coincidences of resonances recorded in Table I may be removed and others introduced.
conclude that the coupled pair of resonances sl/al must be the other a - C H / N H pair of rings VI and VII. The NOE a2 s3 points to a link, through an amide bond (see 20), of the coupled a2/s2 system to ring 11.
-
molecule around rings 111, IV, and V, there are not sufficient data to distinguish the connectivity without considering all the possible models. This is most easily done diagrammatically; to illustrate, the scheme of connectivity for rings I, 11, VI, and VI1 is presented in 22.
+SH
Me0 I
Ar
amide
0 20
22
The carbonyl group bonded to the CY carbon which also bears s2 must provide the link to the ring I NH. This arrangement is evidenced by the proximity of s2 and S6 (Table 111), bearing in mind also that the coupled z/s6 system has previously been shown to be attached to ring I. Thus, the carbonyl group of ring I remains to be attached to a1 (see 5 ) . On the basis of these arguments and model building it follows that the carbonyl group attached to the same carbon as S I cannot be involved in an amide bond; it must therefore be part of the methyl ester function8 of ristocetin A (see 5 ) . If the NOE s2 f is taken to most probably indicate the attachment of the carbon bearing s2 to ring VI, then we arrive at the connectivity scheme shown in 21. This last point in the argument is dubious, since the s2
-
We are fortunately able to limit the possibilities, since it has already been established that rings I1 and I11 bear CYCH-NH groups where the N H is part of a secondary amide. We therefore consider only the structures where either ring IV or V bears the -NH2 group. There are six such structures, three each for ring IV or V bearing the primary amine. Those for ring IV bearing the amine are illustrated in 23-25; the others are obtained by interchanging IV and V.
FoTo23
24
25
-
21
f NOE is shown by models to be consistent with attachment of the s2-bearing carbon to either rings V I or VII. However, it will be seen subsequently that 21 provides a striking analogy to part of the vancomycin structure,I3 and further that there is indeed a close correspondence between the vancomycin12 and ristocetin A chemical shifts and coupling constants for this part of the structure (see later). We have now assigned the aCH-NH pairs for six rings: I, 11, 111, VI, V11, and one of IV and V. The seventh ring must bear the nuclear NH2 group of ristocetin A. Although a number of NOEs have been observed for the region of the
Structure 24 contains two rings neither of which can be constructed using CPK models. Structure 25 is in fact a diketopiperazine, which would require a fairly rigid sixmembered ring. The relevant J u , l \ ; coupling ~ constants are 10 and 9.3 Hz, characteristic of an almost exactly trans-coplanar arrangement of the aCH-H groups, and too large to be accommodated by a cis structure. Thus 23 is the only acceptable scheme of connectivity. Location of, and Stereochemistry Associated with, Rings IV and V The connectivity scheme which emerges from a combination of 22 and 23 (ambiguous only in the possible reverse location
Kalman, Williams
/ Structure of Antibiotic Ristocetin A
903
Table IV. Comparison of Vancomycin and Ristocetin A Chemical Shifts and Coupling Constants (MezSO-ds Solution at 70 "C) ring no.
I
antibiotic
ring protons
R
7.20 7.28
V
II
R
VI
V R V
VI1
R
7.26 7.19
V
7.39 7.48 5.34 5.21 6.84 6.78 6.32 6.30
(7.44) (7.87) 5.85 5.63 6.77 6.73 6.85 6.44
chemical shift values, 6 PCHOH (6.80) (C1)
CYCH
NH
Jn.3
4.38 4.22 5.65 5.7 I 4.73 4.50 4.55 4.50
7.20 6.50 7.66 8.04 8.56 8.43 9.26 8.39
-0 -0
Ja.hn
12 12 8.2 8 6.5 5 5 5
rings I. 11, VI, and VI1 are S, R , R, and S , respectively, and that the absolute stereochemistry a t the center carrying the aliphatic hydroxyl group is R . Particularly convincing in relation to these conclusions is the "nest" of mutual NOEs involving s2, s6, and f; the analogous protons are found in close proximity in vancomycin.13 With the structure of the molecule in the regions of rings I, 11, VI and VI1 defined, we are now in a position to consider the relative locations of rings IV and V. The NOEs for the region of the molecule constituted from rings IV and V have already been listed in Table 111. However, some of the ambiguities can now be removed. Thus a5, already known to be near ring 111 (NOE a as), must be responsible for the N O E (b c a5 a6) q, a4 for the N O E (c a4) q, and cc for the N O E (aa bb cc) h. The simplified list of relevant hOEs is a a5, a3 t, a5 q, p s5, cc h. Also, in the spectra of the antibiotic-peptide complex,I4 we find NOEs corresponding to ( 0 p q) s5. The NOEs p s5 and cc h provide strong evidence for the structure given in 5, i.e., with the free NH2 group on ring IV. There is also clear evidence that the absolute stereochemistry at the asymmetric centers of the second &hydroxytyrosine is the same as that occurring in v a n c o m y ~ i n . The ' ~ NOEs observed in ristocetin A establish that, as in vancomycin, z is on the "front" face of the molecule (the side that binds AC-DA l a - ~ - A l a I ~whereas ), y is on the opposite face (see 10). Moreover, J,,p (the y/s4 coupling constant) is similar in vancomycin (4 Hz) and ristocetin A (5 Hz), as is also J a , (the ~ ~ sq/a5 coupling constant), which is 9 Hz in vancomycin and 9.3 Hz in ristocetin A. We conclude that the absolute configuration at the carbon bonded to s4 is R , and that at the carbon bonded t o y is R . It is now possible to deduce the absolute stereochemistry at the carbon atom bearing s5 ( 5 ) . Since J,,NH for a4/sg is I O Hz, the dihedral angle between the protons of the a C H - N H bond must be close to 180'. If we assume an R configuration at the carbon bearing s5, then for a trans amide bond between rings V and 111 (Le., amide bond involving a4) the conformation of the antibiotic involves a peptide backbone in an extended conformation (27), except for the cis-amide bond indicated (which is known from the X-ray structure of vancomycinI3). Evidence to be presented s ~ b s e q u e n t l yestablishes ~~ that the N H protons a1 and a5 form hydrogen bonds to carbonyl groups of bound Ac-D-Ala-D-Ala. The peptide backbone of the anti-
of rings IV and V) shows that the connectivity of rings I, 11, VI, and VI1 is precisely that found in vancomycin.13 Since both antibiotics bind strongly to peptides terminating in D-AlaD-Ala,'O it is clearly likely that the system associated with rings I, 11, VI, and VI1 provides a common part of a binding site for such peptides. This proposal would be in accord with the proposal for the binding site of v a n ~ o m y c i nSuch . ~ ~ common parts of a binding site would further be expected to be the same in stereochemical detail. For the two antibiotics a comparison (Table IV and 26 (data for vancomycin in parentheses)) of the
+ + + - - - - + + -
OR (7.28)
5.17 5.13
coupling constants, Hz
I
+ +
+
Ho -OHoH 6.85 (6.44)
26 -
chemical shifts and coupling constants for the common parts (where they are likely to be remote from different structural features) shows striking agreement between the two sets of data. With the exception of proton 1 , for the 14 carbon-bound protons which are compared, the chemical shifts are in all cases in agreement to 0.23 ppm or less, with an average deviation of only 0.10 ppm. We conclude that these parts of the two antibiotics are identically constituted (with the exceptions of the chlorine atom in ring I of vancomycin and the conversion of the ring VI1 C02H group in vancomycin to COOMe in ristocetin A), especially in the light of the near identity of the four pairs of J a , values ~ ~ (Table IV). Thus, we conclude that the absolute stereochemistries a t the amino acid a-CH groups of
Cis
27
-
Journal ofthe American Chemical Society
904
OR, Figure 4. Proposed structure of the aglycone of ristocetin A (R, = R 2 = R 1 = H).
biotic does not have the flexibility to make this pssible in this model (27) while keeping the angle between s5 and a4 near to 180’ ( J a p =~ I O Hz observed in the spectrum of the antibiotic-peptide ~ o m p l e x ) . ’ ~ The alternative arrangement has the S configuration at the carbon bearing ss. The J - . N H coupling constant (1 0 Hz) now requires the ring V N H to be on the “front” face of the molecule (28, which shows the major modification caused to the
I
29
right-hand portion of 27). The resulting peptide backbone is bent so that a l and as approach each other more closely, enabling a binding model to be readily c o n ~ t r u c t e d . ’ ~ We conclude that the configuration at the carbon bearing s5 is S . This conclusion is consistent with the NOES p ss, a4 q. as q, and cc h. Thus, the most likely structure for ristocetin aglycone is 5 with the stereochemistry as previously detailed in this paper. The absolute configuration at the ring IV a - C H remains to be determined, although titration data support the S configuration:Is we note that, if it is S as shown in Figure 4, the free NHz of this residue would approach close to the N H as and provide a Coulombic contribution (-CO*--+NH,-) to the binding of Ac-D-Ala-D-Ala as its carboxylate. A CPK model of the proposed structure, given in Figure 5 , shows that a3. a4. and as are in a pocket on the “front” face and so should not be completely accessible to solvent. I n contrast, a ] , although on the front face, should be more accessible (Figure S ) , and a6 and a*, on the “back” face, more accessible still. These expectations are in accord with the measured temperature coefficients of the amide N H proton resonances, which are a3 (-3), a4 (-O.S),as (-1.5);aI (-3),a6 (-5S),az(-6X 10-3ppm/K in Me2SO-d6 solution).
-
from one of the high-field geminally coupled pair eel and eez. Since bb resonates as a broad singlet, the ristosamine anomeric proton must be equatorial. On the basis of the known stereochemistry of ristosamine,16 this may be accommodated i n an a anomer 29 or a anomer 30. Since 30 should show a strong preference to exist in the alternative chair form (with three substituents equatorial and only one axial), we conclude that ristosamine is present as the a anomer 29.
CH,
28
- -
Figure 5. CPK model of ristocetin A aglycone. showing amide N H groups on the “front” face of the molecule. These N H groups are indicated by “hooked” hydragcn atoms, and from left 10 right are a,. a,, a+ and as. The remaining two N H protons (a2 and an) are hidden on the “back” faceof thc molecule and are completely exposed to solvent.
-
U
“back”
/ 102:3 f January 30, 1980
-
Assignment of the Sugar Anomeric Protons and Related Structural Conclusions The partial assignment of the six sugar anomeric protons is given immediately below 5. The ristosamine anomeric proton bb is assigned unambiguously since ristosamine is the only 2-deoxy sugar in ristocetin A, and therefore can be decoupled
OR
30
We have recently prepared a pseudoaglycone from ristocetin A by treatment with MeOH-HCI;” this product has lost five sugars but retains ristosamine. Ristosamine is therefore directly bonded to the aglycone. The resistance to hydrolytic removal of ristosamine by mild acid treatment, and the observation that it may be removed on treatment with base,I8 suggests that it is not linked via a phenolic OH but perhaps via one of the two aliphatic hydroxyl groups of 5. This conclusion would be consistent with the earlier deduction9 that a tetrasaccharide and mannose are separately bound to the aglycone, and yet allow for the observation of four phenolic O H resonances (rather than the three permitted if ristosamine were bound via a phenol) in the present work. Of the six anomeric protons, only one (x) is split by a large coupling constant (7.8 Hz). The anomeric stereochemistry of a trisaccharide unit has been established by the determination of the structure of 0-a-L-rhamnopyranosyI( l - 6 ) - O - [ a - ~ mannopyranosyl( I-2)]-~-glucose;’ this structure has been confirmed by s y n t h e ~ i s . ’The ~ (Y stereochemistries at the anomeric carbons of L-rhamnose and D-mannose preclude large J1.2 coupling constants for probable conformations of either of these sugars. The fourth sugar of the tetrasaccharide is attached at the 2 position of mannose as an 0-0-D-arabinopyranosyl glycoside,20 again excluding a large 51.2 coupling constant for either chair conformation of arabinose, e.&?., 31. The same exclusion applies for the remaining D-mannose monosaccharide unit in a probable conformation of either an a or 0anomer, due to its C-2 stereochemistry (axial substituent). Therefore, x is due to the anomeric proton of D-glucose, and we conclude that the tetrasaccharide unit is bound as its
Kalman, Williams
/ Structure of Antibiotic Ristocetin A OH
905 in Figure 4 represents the tetrasaccharide. The point of attachment of ristosamine, as the cy anomer 29, remains in doubt but it is not bound via a phenolic oxygen. The available evidence would be satisfied, for example, by R3 = ristosaminyl (Figure 4.) A significant proportion of the structural and stereochemical information given in Figure 4 is based on the application of the nuclear Overhauser effect. However, it should be emphasized that the NOE should be applied with care since a single misassigned resonance can give rise to erroneous structural conclusions. In the present work. the possibility of error has been greatly reduced by the correspondence between a portion of the structure of ristocetin A (Figure 4) and that of vancomycin, which has been established by X-ray methods.13
HO O\
0
I
HO 31
anomer at glucose. The present conclusion with regard to the anomeric stereochemistry a t D-glucose is in agreement with those published earlier,9 although the abstract does not provide the evidence for these conclusions. Resonance w is reduced in intensity by -30% on irradiation of the coincident resonances k, 1. This negative NOE is most likely due to the close proximity of the anomeric proton w and an aromatic ring proton ortho to the phenolic oxygen involved in the glycosidic bond. One (I) of the two irradiated protons (k, 1) is indeed ortho to two phenolic oxygens. Since one of these oxygens has already been established to be present as a free OH group, then the remaining phenolic oxygen of ring VI1 must carry the sugar whose anomeric proton is w. On the basis of the above arguments that only the tetrasaccharide and an additional mannose are glycosidically bound via phenolic oxygens, then w is the anomeric proton of the mannose monosaccharide unit (mannose,). The relative sharpness of resonance w is consistent with either an cy- or P-glycoside of mannosel; other workers9 have stated that the sugar is attached to the aglycone as a p-glycoside, although the evidence for this conclusion is not available in the abstract. Specific assignments of the anomeric protons of mannosez, arabinose, and rhamnose (aa, dd, and u) are not made. However, their occurrence as singlets supports the presence of the chair conformations for rhamnose and mannose given in 31. Since the phenolic oxygens indicated by asterisks in 5 are established to exist as free O H groups i n ristocetin A, then the tetrasaccharide unit must be attached to the phenolic oxygen of ring I1 or ring IV. The chemical-shift analogies for H-1 and H-2 of glucose in vancomycin (5.35 and 3.62 ppm) and ristocetin A (5.25 and 3.66 ppm) are consistent with the attachment of the tetrasaccharide to ring 11. Conclusions The structure of the ristocetin A aglycone which is proposed on the basis of the present work is given in Figure 4. The stereochemistry at the cy carbon of ring I V remains in doubt. In ristocetin A, mannose is attached to ring VI1 (Figure 4, R1 = mannosyl). The tetrasaccharide 31 is bonded via the phenolic oxygen of ring I 1 or ring IV, i.e., one of the Rr groups
Note Added in Proof. The remaining details of the structure of ristocetin A have now been established (Williams, D. H.; Rajananda, V.; Bojesen, G.; Williamson, M. J . Chem. SOC., Chem. Commun., 1979,906). Acknowledgments. J.R.K. thanks the University of Sydney for the award of an Eleanor Sophia Wood Travelling Fellowship. We thank Lundbeck (Copenhagen) for generous gifts of ristocetin A . We are grateful to the National Institute for Medical Research, Mill Hill. London, the University of Copenhagen, and the University of Gronigen for generous allocations of time on Brucker 270- and 360-MHz N M R spectrometers. Preliminary experiments were carried out on a Varian XL-100 spectrometer (at the University of Cambridge), purchased with funds provided by the Science Research Council, U.K. References and Notes (1) Williams, D.H.; Rajananda, V.; Kalman, J. R. J. Chem. Soc., Perkin Trans. i 1979,787. (2) Bognar. R.; Sztaricskai, F.; Munk, M. E.; Tamas, J. J. Org. Chem. 1974, 39. 2971. .., -. (3) Sztaricskai, F.; Bognar, R.; Puskas, M. M. Acta Chim. Acad. Sci. Hung. 1975. 84. 75. (4) Lomakina, N. N.; Muravyeva, L. I.; Puskas, M. Antibiotiki(Moscow) 1968, 13, 867. (5) Harris, T. M.; Fehlner, J. R.; Raabe A. B.; Tarbell, D. S. Tetrahedron Lett. 1975, 2655. (6) Harris, C. M.; Kibby, J. J.; Harris, T. M. Tetrahedron Lett. 1978, 705. (7) Fehlner, J. R.; Hutchinson, R . E. J.; Tarbell, D. S.; Schenk, J. R., Proc. Natl. Acad. Sci. U.S.A. 1972, 69, 2420. 181 Boanar. R.: Sztaricskai. F.: Brazhnikova. M. G.: Lomakina. N. N.: Zenkova. V. 'A. Magy. Kem. Foly. 1974, 80, 503. (9) Lomakina, N. N.; Bognar, R.; Brazhnikova, M. G.; Sztaricskai, F.; Muravyeva, L. I. Abstracts, Seventh International Symposium on the Chemistry of Natural Products, Zinate, Riga, 1970, p 625. (10) Nieto, M.: Perkins, H. R . Biochem. J. 1971, 124, 845. See also: Perkins, H. R. /bid. 1969, 1 1 1, 195. (11) Gale, E. F.; Cundliffe, E.; Reynolds, P. E.; Richmond, M. H.; Waring, M. J. "The Molecular Basis of Antibiotic Action": Wiley, New York. 1972, pp 102-1 13. . (12) Williams, D. H.; Kalman, J. R. J. Am. Chem. SOC. 1977, 99, 2768. For further considerations of the negative NOE, employed in this paper, see: Gibbons, W. A.; Crepaux, D.; Wyssbrod, H. R., "Proceedings of the 4th American Peotide SvmDosium". Walter, R.. Meinhofer. J.. Eds.: Ann Arbor Press: New York. 1975. (13) Sheldrick, G. M.: Jones, P. G.: Kennard. 0.: Williams, D. H.; Smith, G. A. Nature (London) 1978, 271, 223. (14) Kalman, J. R.; Williams, D. H. J. Am. Chem. Soc., following paper in this issue. (15) Rajananda, V.; Williams, D. H., unpublished work. (16) Lee, W. W.; Wu, H. Y.; Marsh, J. J.; Mosher, C. W.; Acton, E. M.; Goodman, L.; Henry, D. W., J. Med. Chem. 1975, 18, 767. Sztaricskai, F.: Pelyvas, I.; Bognar, R.; Bujtas, Gy. Tetrahedron Lett. 1975, 11 11 (17) Norris, A. F.; Williams, D. H., unpublished work. ( 1 8 ) Lomakina, N. N.; Spiridonova, I. A,; Bognar, R.; Puskas, M.; Sztaricskai, F. Antibiotiki (Moscow) 1968, 13, 975. (19) Sztaricskai, F.; Liptak, A.; Pelyvas, I. F.; Bognar R. J. Antibiot. 1976, 29, 626. (20) Neszmeiyi, A.; Sztaricskai, F.; Liptak, A,; Bognar, R. J. Antibiot. 1978, 31, 974. I
/
~~